Increasing Lentiviral Transduction Efficiency: Towards Cost- Effective T Cell Therapy Manufacturing Benjamin A. F. Blaha1, Alexia Toufexi1, Gregory Berger, Enas Hassan, and Nicholas R. Gaddum

Cell and Catapult, 12th Floor Tower Wing, Guy's Hospital, Great Maze Pond, London, SE1 9RT, United Kingdom 1 Introduction 3 Results Challenge: High-cost manufacturing processes of Chimeric Antigen Receptor (CAR) T-cells therapies are prohibitively expensive (Fig. 1) [1,2]. Due to the cost of , transduction is a major cost driver in Time course profiles CAR T-cell manufacturing. Several bioprocessing parameters have been identified as potentially playing a role in transduction efficiency, such as the physical proximity of lentivirus particles to T cells. This Variable Agitation proximity could be manipulated through (1) the number of cells and virus particles in the suspension; (2) Total number of transduced CD3+ cells at variable agitation Total number of CD3+ cells at variable the periods of agitation to encourage homogeneity; and (3) the surface-to-volume ratio in the agitation

transduction vessel [3]. However, limited research has been performed on identifying and optimising ) 6

critical process parameters of transduction. ) 6 Aim: Investigate sensitivities of critical bioprocess parameters upon T-cell transduction. Manufacturing Process of CAR T-cell Therapy

T-Cell1 isolation Total CD3+ cells (x10

Time (Days) T-Cell Engineering Leukapheresis Figure 4. Transient, total number of Total transduced Total transduced CD3+ cells (x10 CD3+ and transduced CD3+ cells through: no agitation; infrequent agitation; and Time (Days) CAR T-Cell frequent agitation. Infusion • Agitation has a negative effect on transduction efficiency and the total number of transduced cells • 2.6-fold higher transduction efficiency at static cultures CAR T-Cell Expansion • 6.9-fold difference of total transduced cells between static and frequently agitated cultures Figure 1. Schematic representation of the manufacturing process of CAR T-cell therapies. Cells to surface area ratio

Percentage of transduced CD3+ cells at variable cells to Total CD3+ cells at variable cells to surface area ratio surface area ratio

Methods ) 2 6 A cryopreserved CD3+ T-cell bank generated from fresh leukapheresis was used. Over a period of four days, transduction of cells was achieved in 36 parallel culture conditions in wells using GFP lentiviral vectors (MOI=1). Investigated process parameters included (1) working volume (1-3 mL); (2) agitation

frequency (0-8 cycles/day); (3) cell seeding concentration (0.5-1.0 viable cells/mL), and (4) activation Total Total CD3+ cells (x10 agent (0.5-1.5 U/mL) (TransActTM, Miltenyi Biotec, Germany). Agitation was performed using orbital

shakers at 500 rpm at a different number of frequency cycles of 30 min intervals per day. After cell Time (Days) % Transduced % Transduced CD3+ cells counting and flow cytometry analysis, time course profiles and multivariate screening models were generated (Fig. 2). Figure 5. Percentage of transduced CD3+ and total CD3+ cells at variable cell to Bioprocessing Transduction Expansion Data Analysis surface area ratios. Parameters Time (Days) • Increasing the cells to the surface area ratio has a negative effect on transduction efficiency Activation • 2.0-fold higher levels of transduction efficiency between low and high volume cultures (0.5-1.5 U/mL) Cell Counts Working Multivariate screening profiles Volume (1-3 mL) Seeding density Flow Table 2. Significant terms (p<0.05) for screening models 1-7. For each response function, this table shows GFP (0.5-1.0 viable Cytometry normalised gradients (∇) to each significant model term. Factors: Seeding volume (Vi), agitation frequency CD3+ Lentiviral Vector cells/mL) (N), viable cell seeding density (VCD ), and TransActTM concentration ([TA]) and factor interactions [4]. T-cells (MOI = 1) 0-4 days i Agitation Significant model terms Responses (0-8 cycles/day) Vi N VCDi [TA] Vi N Vi VCDi N VCDi Vi N·VCDi

Figure 2. Schematic representation of the experimental plan. ∇ 0.4789 -0.0895 0.2444 0.1159 1 CD3TOT p < 0.0001 0.0012 < 0.0001 0.0001 Table 3. Relative contribution of Time course profiles ∇ 0.0053 -0.0057 0.0075 0.0023 variance for each individual studied 2 CD3/LIVE (%) p < 0.0001 < 0.0001 < 0.0001 0.0349 factor. One-Factor-a-Time (OFAT) ∇ 0.1574 -0.0976 0.1582 -0.0805 0.0636 -0.0350 Factors Responses 3 CD3, GFP+TOT p < 0.0001 < 0.0001 < 0.0001 0.0001 0.0009 0.0320 Vi N VCDi [TA] CD3 74% 3% 23% 1% TransActTM=1%, ∇ -0.0612 -0.0347 0.0928 -0.0694 TOT TransActTM=1%, GFP+/CD3 VCD=0.75 cells/mL, 4 CD3/LIVE (%) 46% 30% 5% 18% VCD=0.75 cells/mL, (%) p < 0.0001 0.0014 < 0.0001 < 0.0001 Infrequent Agitation CD3, GFP+TOT 42% 20% 36% 1% V=2 mL ∇ 1.3959 -1.5875 1.4206 -0.9652 GFP+/CD3 (%) 37% 25% 37% 1% (4 cycles/day) 5 Virus:GFP+ p 0.0002 < 0.0001 0.0002 0.0038 Virus:GFP+ 26% 40% 30% 4% Infrequent Frequent V=1 mL, V=2 mL, V=3 mL, No Agitation • The lowest virus cost per transduced cell was achieved at (1) no agitation, (2) high initial cell Agitation Agitation O.75 x 106 / 9.5 1.5 x 106 / 9.5 2.25 x 106 / 9.5 (0 cycles/day) concentration, and (3) high culture volume. This corresponded to a 4.6-fold cost-reduction relative to the (4 cycles/day) (8 cycles/day) (cells/cm2) (cells/cm2) (cells/cm2) highest observed virus cost per transduced cell (3.0 vs. 13.8 per transduced cell).

Figure 3. Schematic diagram showing the different conditions investigated in the OFAT. 6 6 VCDi=0.5 x 10 cells/mL VCDi=1.0 x 10 cells/mL Multivariate screening profiles

Design of Experiments (DoE)

Design Space Factors -1 0 1 Table 1. Design space for a 2-level 4- Figure 6. Fitted response factor full factorial DoE screening surface model for virus cost Volume (mL) 1 2 3 experiment, in which controlled variation per transduced cell due to

TM Cost (Virus/GFP+ cell) Agitation of agitation, VCDi, volume, and TransAct variation of agitation cycles 0 4 8 (no 30 min intervals/day) concentration was introduced. All (N) and working volume (V) statistical analysis and model generation at (A) low seeding density [TransAct] (% vol.) 0.5 1 1.5 N (int/day) N (int/day) was performed using Design Expert (Stat- V (mL) V (mL) (VCDi) and (B) high seeding 6 VCDi (10 /mL) 0.5 0.75 1 Ease Incorporation; Minneapolis, MN). density. 4 Discussion & Conclusions ❑ Transduction efficiency, total transduced cells and cost per transduced cell were significantly sensitive to variations in agitation, culture volumes and seeding density. ❑ High cell seeding densities and lower working volumes, with little to no agitation allows maximum transduction of T cells, perhaps due to congregation of cells and virus. ❑ Virus cost per transduced cell can be dramatically lowered through bioprocess optimization.

References 1. Levine BL, Miskin J, Wonnacott K, Keir C. Global Manufacturing of CAR T Cell Therapy. Mol Ther - Methods Clin Dev. 2017;4(March):92-101. doi:10.1016/j.omtm.2016.12.006. 2. James D. How short-term gain can lead to long-term pain. Cell & Gene Therapy Insights. 2017;(April):271-283. doi: 10.18609/cgti.2017.018. Cell and Gene Therapy Catapult 3. Verhoeyen E, Costa C, Cosset F-L. Lentiviral Vector Gene Transfer into Human T Cells. In: ; 2009:97-114. doi:10.1007/978-1-59745-409-4_8. 12th Floor, Tower Wing, Guy’s Hospital, Great Maze Pond, London SE1 9RT 4. Anderson M., Whitcomb P. DOE Simplified: Practical Tools for Effective Experimentation, 3rd Edition, CRC Press. +44 (0) 203 728 9500 | [email protected] | ct.catapult.org.uk

Cell and Gene Therapy Catapult is a trading name of Cell Therapy Catapult Limited, registered in England and Wales under company number 07964711.